Objective: To determine by meta-analysis the efficacy of mood stabilizers i
n preventing recurrence of bipolar or unipolar mood disorders and to consid
er the evidence for a lithium withdrawal-induced relapse syndrome.
Method: Controlled studies of lithium, valproate and carbamazepine in preve
nting future episodes of affective disorders were classified according to m
ethodological rigour, and mete-analyses were performed overall and on each
type.
Results: A total of 19 blinded, randomized, controlled trials of prophylaxi
s in 865 patients found lithium highly effective (74% recurrence on placebo
vs. 29% on lithium). In the mirror-image studies, whose substantial lithiu
m vs. prior treatment difference cannot be explained by withdrawal relapse,
lithium reduced relapse by 50% (bipolar) and 58% (unipolar).
Conclusion: Maintenance lithium produces a highly significant reduction in
relapses. The mirror-image studies had not been systematically analysed pre
viously, and they support the effectiveness of lithium. We also failed to f
ind sufficient evidence to prove that the lithium-withdrawal relapse phenom
enon exists.